LabelD is a quick and easy-to-use image annotation tool, built for academics, data scientists, and software engineers to enable single track or distributed image tagging. LabelD supports both localized, in-image (multi-)tagging, as well as image categorization.
LabelD was created as a simple image annotation tool to minimize the amount of work/time spent on annotation by streamlining the overall process. The images can either be pulled from Imgur based on keyword search (search button at the top right), or locally (please un-select Imgur as the image source under Annotation Settings).
LabelD is simple enough to get up and running.
And... you're done!
If you dont have an image base to start from, this may be the easiest platform to begin building a training set. It comes default with the ability to pull keyword indexed images from Imgur, so try it out!